HIGH ENERGY FLAT PANEL PORTAL IMAGER
Field This patent specification is in the field of imaging high-energy radiation, such as used in radiation treatment of patients, to keep track of the path of high-energy radiation relative to the patient's anatomy.
Background High-energy radiation, in the mega electron volt range, such as 6 MEV, has long been used for oncology and other treatment of patients. In principle, high-energy treatment proceeds in accordance with a treatment plan that aims to deliver a high dose of the high-energy radiation at a lesion but a much lower dose elsewhere in the patient's body. Typically, there is relative motion between the source of the high- energy radiation and the patient such that a high-energy beam always passes through the lesion being treated but has different paths through the patient's body at different times. For example, radiation treatment systems that include a linear accelerator are available under the tradenames Clinac from Varian and Oiicor from Siemens. Information on such systems is available at websites such as <varian.co/onc>, <siemensmedical.com>, and <topdocsonline.com/radiation_oncology>, and is hereby incorporated by reference in this patent specification. It is desirable to know the path of the high-energy beam through the patient's body in order to ensure that it passes through the lesion being treated. However, the high energy of the treatment beam makes it difficult to use conventional radiation imaging devices. One known approach is to position that patient, open the aperture of
the radiation source and emit a short burst of high-energy radiation that is imaged with an x-ray receptor placed under the patient. The resulting image is used to control a shutter and/or collimator at the source, and then another burst of radiation is emitted in a narrow beam that passes through the lesion being treated. The source (and/or the patient) then moves to a number of different positions, and the procedure is repeated at each new relative position between the radiation source and the patient, until the desired cumulative radiation dose has been delivered to the lesion. The imaging used in this process typically is called '"portal imaging." Flat panel x-ray imaging receptors or imagers have recently become widely used in diagnostic imaging. Some use indirect conversion in which x-ray energy is converted to light energy and then the light energy is converted to electronic information using an array of photodiodes and transistors. Others, such as those commercially available from the assignee of this patent specification, Hologic, Inc. of Bedford, MA, use direct conversion of x-ray energy to electronic information, using a Selenium-based plate and an array of transistors. For example, it is believed that the current Siemens Oncor Linear Accelerator uses an indirect conversion flat panel imager for portal imaging. Portal imaging with direct conversion flat panel imagers can be done in a way that reduces undesirable effects on imaging due to the use of high-energy radiation in radiation treatment systems. One such undesirable effect is ghosting that involves residual effects on an image caused by previous imaging of high-energy radiation. See G. Pang, D.L. Lee, and J. A. Rowlands, "Investigation of a direct conversion flat panel imager for portal imaging," Med. Phys. 29 (10), 2121-28, October 2001, which
is hereby incorporated by reference. As noted therein, ghosting can appear in both indirect conversion and direct conversion panels. Particularly when using a direct conversion imager, it is desirable to know the response (sensitivity) of each picture element (pixel). With such knowledge, the pixel values for a portal image can be adjusted before the image is displayed or is otherwise used. For example, if pixel A has a sensitivity of 2 relative to pixel B, then after a portal imaging x-ray exposure the reading from pixel A can be divided by 2 before the image is displayed or otherwise used. In principle, pixel sensitivity information can be obtained by exposing the imager to a flat or known field x-ray beam to generate a calibration file used to correct the raw pixel readings in a later x-ray image. Ghosting occurs when the sensitivity of pixels has changed due to previous x- ray exposures. If, for example, a small section of the imager has been irradiated, the pixels in that section typically will have altered sensitivity. The same pixel A, if in the irradiated section, may now have a sensitivity of 1.5 relative to pixel B. If the calibration file is not changed, when a subsequent reading is taken for another x-ray image, the reading for pixel A will be divided by 2 instead of by the correct factor of 1.5. The displayed intensity of pixel A will therefore be too low. The ghost of the previous x-ray exposure(s) will therefore appear in the new image due to the altered sensitivities of pixels that were included in the previous exposure(s). It is believed that improvements can be made in portal imaging, and this patent specification is directed to such improvements.
Summary This application provides a system for radiation treatment of patients which, according to an exemplary embodiment comprises: a source of high-energy radiation, a digital flat panel imager and a computer storing a light reference image derived from the imager when rested; the computer being programmed to derive from the digital flat panel imager (i) a light testing image in response to exposure to light, after exposure of the digital flat panel imager to high-energy radiation, and (ii) a portal image in response to exposure to high-energy radiation from the source; the computer further being programmed to calculate a gain correction using the light reference image and the light testing image; and the computer further being programmed to use the gain correction to correct the portal image for ghosting. A system for radiation treatment of patients, according to another exemplary embodiment, comprises: a source of high-energy radiation, a patient table, and a digital flat panel imager at the other side of the patient table from the source, and a computer storing a light reference image derived from the imager when rested; the computer being programmed to derive from the imager a light testing image in response to exposure to light at a time to, after exposure of the imager to high-energy radiation, and a portal image in response to exposure to high-energy radiation from the source at time t; the computer further being programmed to calculate a gain correction using
the light reference image, the light testing image, and a factor accounting for changes in sensitivity or gain of the imager in an interval between the times to and t as well as a factor accounting for different response or sensitivity of the imager to light from that to x-rays; and the computer further being programmed to use the gain correction to correct the portal image for ghosting due to exposure of the imager to high-energy radiation before the time t0. This application also provides a method for correcting for ghosting in portal imaging with high-energy radiation and a digital flat panel imager. According to an exemplary embodiment, the method comprises: exposing the digital flat panel imager to light to derive a light reference image; exposing the digital flat panel imager to light to derive a light testing image; calculating a gain correction, using the light reference image and the light testing image; exposing the digital flat panel imager to high-energy radiation to derive a portal image; and using the gain correction to correct the portal image for ghosting. A method, according to another exemplary embodiment, for correcting for ghosting in portal imaging with high-energy radiation and a digital flat panel imager, comprises: exposing the imager to light to derive a light reference image; exposing the imager to high energy radiation before a time t0; exposing the imager to light at time t0 to derive a light testing image; calculating a gain correction for a portal image to be taken at a time t, using the light reference image, the light testing image, and a factor accounting for changes in sensitivity or gain of the imager in an interval between the times to and t as well as a factor accounting for different response or
sensitivity of the imager to light from that to x-rays; exposing the imager to high- energy radiation at the time t to derive the portal image; and using the gain correction to correct the portal image for ghosting due to the exposing the imager to high-energy radiation before the exposure at the time t. This application also provides a radiation treatment method. According to another exemplary embodiment, the method comprises: deriving a portal image by exposing a patient to high-energy radiation and imaging the radiation with a digital fiat panel imager; correcting the portal image for ghosting due to previous exposure to high-energy radiation by modifying the portal image in accordance with information regarding a response of the imager to light and a factor related to different responses of the imager to light and to high-energy radiation, to thereby derive a corrected portal image; and using the corrected portal image to guide radiation treatment of the patient. A radiation treatment method, according to another exemplary embodiment, comprises: deriving a portal image by exposing a patient to high-energy radiation and imaging the radiation with a digital flat panel imager; correcting the portal image for ghosting due to previous exposure to high-energy radiation by modifying the portal image in accordance with information regarding a response of the imager to light and at least one of a factor related to decay of images of high-energy radiation taken with the imager and a factor related to different responses of the imager to light and to high-energy radiation, to thereby derive a corrected portal image; and using the corrected portal image to guide radiation treatment of the patient.
Brief description of the drawing Fig. 1 illustrates main elements of a radiation treatment system incorporating a flat panel imager according to a preferred embodiment. Fig. 2 illustrates a flat panel imager for use in the system of Fig. 1. Figs. 3a-3d illustrate images before and after correction in accordance with preferred embodiments. Fig. 4 shows a flow chart illustrating a method for correcting for ghosting in portal imaging with high-energy radiation and a digital flat panel imager, according to an exemplary embodiment. Fig. 5 shows a flow chart illustrating a radiation treatment method, according to an exemplary embodiment. Fig. 6 shows a flow chart illustrating a method for correcting for ghosting in portal imaging with high-energy radiation and a digital flat panel imager, according to a preferred embodiment. Fig. 7 illustrates a timing relationship. Fig. 8 shows a flow chart illustrating a method for correcting for ghosting in portal imaging with high-energy radiation and a digital flat panel imager, according to another exemplary embodiment. Fig. 9 shows a flow chart illustrating a radiation treatment method, according to an exemplary embodiment.
Detailed description of preferred embodiments Portal imaging with direct conversion flat panel imagers can be used to
advantage by techniques that reduce undesirable effects on imaging due to the use of high-energy radiation in radiation treatment systems. One such undesirable effect is ghosting that involves residual effects on an image caused by previous imaging of high-energy radiation. In accordance with preferred embodiments discussed in more detail below, additional corrections are introduced in processing the image information from a direct conversion flat panel imager to account for portal imaging applications in high- energy radiation treatment systems. In particular, in addition to other calibrations, the energy conversion layer of the imager is illuminated with light energy to produce an image that contains information related to ghosting and other effects due to previous imaging of high-energy radiation, and this image is used in correcting a portal image of high-energy radiation that has passed through a patient being treated. Such use of an image of light energy is particularly effective in direct conversion light panels because they respond effectively to illumination with both light energy and x-ray energy (which may not be the case with indirect conversion panels), and because of generally better localization of interactions between high- energy x-ray photons and energy conversion materials. Information useful to account for ghosting can be obtained with the patient in place, between exposures to high-energy treatment radiation. Fig. 1 schematically illustrates main elements of a radiation treatment system 1 such as those incorporated by reference above, with the exception that the flat panel imager and its interaction with the rest of the system are different and contribute to the improvements described below. The system 1 includes a table 10 on which the
patient (not shown) is positioned, a source 12 of high-energy radiation that includes a conventional shutter and/or collimator to control a radiation beam 14 emitted thereby, a position control 16 that moves at least one of the source 12 and patient table 10 relative to the other, a novel flat panel imager that is at the other side of the patent from source 12 and is positioned to receive radiation beam 14 after it has passed through the patient, and a computer control 20 that includes conventional operator interface components and controls the operation of the system, and also novel components that interact with flat panel imager 18. Except for the use of the novel flat panel imager 18, radiation treatment with the system of Fig. 1 can be similar to using a conventional oncology radiation treatment system of the type referred to above. As discussed below, the computer 20 is preferably programmed to derive from the imager 18 (i) a light testing image in response to exposure to light, after exposure of the digital flat panel imager to high-energy radiation, and (ii) a portal image in response to exposure to high-energy radiation from the source. The computer is further programmed to calculate a gain correction using the light reference image and the light testing image, and to use the gain correction to correct the portal image for ghosting. The imager 18 preferably comprises a direct conversion imager. The imager may comprise an internal light source selectively energized to expose the imager for deriving the light reference and light testing images. The internal light source is preferably at a side of the digital flat panel imager facing the source of high energy radiation. The imager preferably comprises a Selenium-based layer for converting light energy and high-energy x-rays to electrical signals.
As discussed below, according to another preferred embodiment, the computer 20 is preferably programmed to derive from the imager a light testing image in response to exposure to light at a time to, after exposure of the imager to high-energy radiation, and a portal image in response to exposure to high-energy radiation from the source at time t. the computer is further programmed to calculate a gain correction using the light reference image, the light testing image, and a factor accounting for changes in sensitivity or gain of the imager in an interval between the times to and t as well as a factor accounting for different response or sensitivity of the imager to light from that to x-rays, and to use the gain correction to correct the portal image for ghosting due to exposure of the imager to high-energy radiation before the time t0. Fig. 2 illustrates a vertical section (not to scale) through flat panel imager 18 that is a modified version on an imager commercially available from Hologic, Inc. under the tradename DirectRay direct-to-digital image capture detector. Principles of the construction and operation of such an imager are discussed in commonly assigned U.S. Patent No. 5,319,206, which is hereby incorporated by reference. The commercially available DirectRay imager is modified for the purposes disclosed in this patent specification in several ways. The top cover 22 of imager 18 is raised so that it is spaced further from a top electrode 24 of the imager as compared with the commercially available imager, and a buildup plate 22a of x-ray attenuating material, such as a 2 mm thick copper plate, is embedded or otherwise affixed to top cover 22. For example, copper plate 22a can be rectangular in plan view, 30 by 40 cm in size.
A light source 26 is placed between top cover 22 and top electrode 24, and can be selectively energized to provide a light exposure that preferably is fiat field exposure, at least over an area of the top electrode 24 that is irradiated with x-ray beam 14, but preferably over a greater area, and most preferably over most or all of top electrode 24. For example, light source 26 can be energized for a period of 0.1 sec, with a power density of 45 microwatt per square centimeter at top electrode 24. However, different period and power densities can be used that provide a suitable light image. If the light exposure is not flat field, an additional correction should be made to account for this. Imager 18 includes a Selenium-based conversion layer that is under top electrode 24 and other layers that are substantially transparent to light. A similar light source (not shown) can be placed at the bottom of imager 18 to illuminate the same Selenium-based layer through other layers that are under it. The two light sources can be operated together or in a desired sequence, or only one of them can be used. The light can be white light, or different light at one or more wavelengths or wavelength ranges that generate a suitable electrical image output from imager 18. A power supply and operator console unit 28 provides power to imager 18, including to light source 26, and otherwise interacts with imager 18 to obtain therefrom electronic imaging information. The conventional parameters for imager 18 preferably are changed to reduce gain to about one-quarter that for a medical grade imager typically used for medical diagnostic purposes, such as in a chest x-ray machine, and to reduce the operating voltage, e.g. to an operating voltage of the order of 2 KV. During normal operation of a direct conversion flat panel imager such as those
commercially available from the assignee hereof, reference image (dark image) data is being acquired continuously and the latest data set, Ref(x,y), is stored in a reference data buffer for each pixel at position (x,y) in the panel. This data set provides a pixel- to-pixel baseline offset information, as the differences in pixel data in response to a known (e.g., fiat) illumination, or no illumination, can be used to correct for such differences when an image of a patient, or an image with different illumination, is taken. The pixel-to-pixel gain (sensitivity) in the case of such a direct conversion flat panel imager can be calibrated by two methods. (1) using x-ray energy, and (2) using internal (erase) light. Medical panels used for general radiography normally can use only method #1. Image data, XG(x,y), from a flat field x-ray exposure are acquired during an x-ray calibration procedure. A gain map GM(x,y) is then generated according to the following equation, where AVG is the average pixel value of the entire panel: GM(x,y) = AVG (Equation 1 ) XG(x,y) - Ref(x,y)
For general radiography imaging, exposure data Exp(x,y) is acquired by x-ray exposure penetrating through the patient. The final image IMG(x,y) is derived according to the following equation: IMG(x,y) = Exp(x,v) - RefTx.v) (Equation 2) GM(x,y)
Gain calibration using x-ray energy normalizes both pixel-to-pixel variations in the panel and intensity non-uniformity of the x-ray profile such as the so called "heel effect," where the x-ray intensity away from the center is lower due to the
increase of x-ray path length and the inverse square law. In portal imaging, the panel can be exposed to extremely high doses of radiation and the sensitivity of selenium which has a history of high doses of radiation can exhibit "fatigue" for a period of time, and a lower sensitivity to radiation until it is recovered. A negative image of the fatigued area can modulate the intensity of sequence images and produce a "ghost image". In accordance with preferred embodiments suitable for use of a direct conversion panel to portal imaging, a new method reduces the effect of this "ghost image". For example, the following procedure may be used: (1) with a well rested imager, so that any ghost images have dissipated, take a reference image, typically without either light or x-ray illumination (a dark current image) to generate and store image data Ref(x,y) for the respective pixel positions at coordinate points x,y; (2) with the imager still well rested, take a substantially flat or known field image with x- rays or with light to generate and store image data XG(x,y); (3) calculate gain map data GM(x,y) = [XG(x,y) - Ref(x,y)]/AVG, where AVG is an average or mean of the pixel values of the entire imager for data XG(x,y); (4) just before taking an x-ray portal image of the patient on the table, take a flat or known field light image to derive and store image data LG(x,y); (5) take a portal x-ray image of the patient to derive and store image data Exp(x,y); (6) calculate image data IMG(x,y) = [Exp(x,y) — Ref(x,y)]/ [GM(x,y)*LM(x,y)], where LM(x,y) = [LG(x,y) - Ref(x,y)]/AVG(light), and AVG(light) is an average or mean of the pixel values in LG(x,y); and repeat steps 2-6 for additional portal images. Image data IMG(x,y) defines an image corrected for ghosting, and can be used to guide radiation treatment as is known in the art. The
variable AVG and AVG(light) can be used in the numerator of the respective expressions rather than in the denominator. When a certain area of the panel is fatigued, and before the next patient image is acquired, a set of panel data LG(x,y) can be acquired using an internal light source(or sources) in the panel. The patient is then exposed to x-ray radiation to obtain image data, Exp(x,y). The ghost image corrected data then is derived in accordance with the following equation, where AVG(light) is the average pixel value of the panel using the internal light source(s): IMG(x,y) = Exp(x,y) - Ref(x,v) (Equation 3) GM(x,y) * LM(x,y)
and LM(x,y) = AVG(Ii ghf) (Equation 4) LG(x,y) -Ref(x,y)
Since LG(x,y) can be obtained without removing the patient, this ghosting removing technique can be repeated as often as needed to minimize the visual effect of ghosting with the patient in place. In operation, with imager 18 well rested so ghost images have been dissipated, images Ref(x,y) are derived as in medical use, and the latest one is always stored in a buffer in computer 20. With imager 18 still well rested, a flat field x-ray exposure is taken to produce image data XG(x,y). The flat field preferably is radiation from source 12, with no patient in place, so that non-uniformities in the "flat field" from source 12 can be taken into account in calculating the gain map GM(x-y). Alternatively, the flat field can be light energy, e.g. from light source 26. Image data XG(x,y) are supplied to computer 20, which calculates gain map GM(x,y) in
accordance with equation (1) and stores the gain map. Then, with the patient in place on table 10, and at a selected time relative to the taking of a portal image of high- energy radiation that has traversed the patient, e.g., just before a portal image is to be taken, image data LM(x,y) is taken by energizing light source 26 and reading out the resulting image from imager (panel) 18 and supplying the image to computer 20 for storage therein. Then, computer 20 commands source 12 to emit an x-ray beam 12 for portal imaging, and image data Exp(x,y) is read out of imager 18 and stored in computer 20, which calculates ghost-corrected image data IMG(x,y) in accordance with equations (3) and (4), and uses IMG(x,y) for treatment planning as known. The process of obtaining and using images LG(x,y), LM(x,y), and IMG(x,y) can be repeated in the course of radiation treatment of a patient while the patient remains on table 10. Thus, although imager 18 may be exposed to high-energy portal imaging radiation on a succession of occasions, effective corrections for ghosting can be made repeatedly using repeatedly taken light images LG(x,y). The corrected image IMG(x,y) derived as described above can be used in place of the conventional portal image in determining the settings of the shutter/collimator of source 12 required for directing a narrow radiation beam along a path passing through the lesion being treated. As is known, this involves determining the location of the lesion in the image based on the image from the imager, in this case IMG(x,y), and using information about the location of the lesion in the image and the position of source 12 relative to imager 18 and/or table 10 to determine the correct beam path. Figs. 3a-3c illustrate experimental results indicative of benefits of the new
approach. Fig. 3 a represents an image of radiation from source 12 that has passed reached the imager directly (with no patient or phantom in the path), but after imager 18 has taken such images previously, so that undesirable ghosting appears. Fig. 3b shows the same image as in Fig. 3 a but corrected as described above. Fig. 3 c represents an image taken under conditions similar to those for Fig. 3 a but with a phantom in the path of the radiation, and Fig. 3d is the same image after correction according to the process described above. A method, according to another exemplary embodiment, for correcting for ghosting in portal imaging with high-energy radiation and a digital flat panel imager will now be discussed with reference to Fig. 4. The digital flat panel imager is exposed to light to derive a light reference image (step S41). A light testing image is also derived by exposing the digital flat panel imager to light (step S43). A gain correction is calculated using the light reference image and the light testing image (step S45). When the digital flat panel imager is later exposed to high-energy radiation to derive a portal image (step S47); the gain correction is used to correct the portal image for ghosting (step S49). Exposing the imager to high-energy radiation to derive the portal image can comprise a direct conversion to electronic output from the imager. At least one of the exposures to light to derive the light reference image and the light testing image may comprise exposure to a flat field light. According to another exemplary embodiment (Fig. 5), a radiation treatment method comprises (i) deriving a portal image by exposing a patient to high-energy radiation and imaging the radiation with a digital flat panel imager (step S51), (ii) correcting the portal image for ghosting due to previous exposure to high-energy
radiation by modifying the portal image in accordance with information regarding a response of the imager to light and a factor related to different responses of the imager to light and to high-energy radiation, to thereby derive a corrected portal image (step S53), and (iii) using the corrected portal image to guide radiation treatment of the patient (step S55). Deriving a portal image may comprise using a direct conversion imager to derive the portal image. Using the corrected portal image may comprise selecting a relative position between a patient and a source of high-energy radiation in accordance with the corrected portal image. The method may further comprise correcting in accordance with a light image taken with the imager when the imager is substantially free of ghosting due to exposure to high-energy radiation. More generally, this patent specification describes tools (in the form of methods, systems and apparatuses) for reducing ghosting in a flat panel imager that images high-energy radiation that has passed through a patient, by means of exposing the imager to radiation different from the high-energy radiation and using image information derived in response to such different radiation to correct image information derived in response to high-energy radiation to reduce undesirable ghosting effects. More specifically, the new method and apparatus involve using such corrections with imagers that directly convert x-ray energy to electronic signals rather than first converting x-ray energy to light energy. Still more specifically, the novel method and apparatus involve conversion utilizing a Selenium-based layer. The different radiation used in the correction can be light of a wavelengths in the range of wavelengths that produce an imaging output from the imager. The tools may be embodied (at least in part) in a computer program stored on
a computer readable medium and/or transmitted via a computer network or other transmission medium. As discussed in more detail below, two additional factors are preferably introduced in calculating an image corrected for ghosting. One of them takes into account the time delay between taking a flat or known field light image at time to and the taking of the immediately following x-ray portal image at time t, and effects of that time delay on the gain map, to thereby improve the ghosting correction. The other factor takes into account the difference between the response or pixel sensitivity of the imager to x-rays and that to light, and effects of that difference on the gain map, to thereby further improve the ghosting correction. hi an exemplary embodiment, a ghosting correction comprises the following steps of using a digital imager in a radiation treatment system of the types discussed above: 0. First we assume that an initial calibration using x-rays has been done, as known in the art, for the imager, as outlined in the background section. That is, (i) with the imager well rested so that any ghost images have dissipated, take a substantially flat or known field image with x-rays to generate and save an initial gain map data GM(x,y); and (iϊ) take dark current images at frequent intervals at times when the imager is not being exposed to light or x-rays to generate dark current images Ref(x,y), and store the most recent image Ref(x,y). Any subsequent image taken with the imager will be automatically subtracted by the currently stored dark current image Ref(x,y) and divided by the initial gain map data GM(x,y) BEFORE the further corrections described below;
1. With the imager well rested, so any ghost images have dissipated, take and save an image by flashing a light that can be built into the imager. This image is Imagehght
jefeience, and includes respective values for the pixel positions in the imager; 2. After taking an x-ray image, say the (N-l)th x-ray image, where N=2,3,..., take and save another light image at time t
0. This image is Image(N-i)th_hght te
stmg(to), and also includes respective values for the pixel positions; 3. Calculate a gain correction G
gi
lost,
ng(t) for a time t, at which the next (i.e., Nth) x- ray image will be taken. The gain correction G
gi
10Sting(t) also comprises respective values for the pixel positions. The calculation is according to:
l,ght
jestmg(t
θ)/
| normalized to 1 at a iefeieiice pixel ~ M 6XP_~*-Λt-to_)J where C is a known constant, derived by testing the imager to determine a rate of decay of a ghost image on the imager. (For example, a part of the imager, say the left half, can be exposed to a substantial amount of radiation, i.e., a ghosting exposure to create ghosting. Then take a number of testing images in a time sequence with the whole imager exposed to a small amount of radiation, i.e., a testing exposure of a flat field each time. From these testing images, the relative difference between the pixel sensitivity in the left half region and that in the right half region can be measured and plotted as a function of the time delay between the ghosting and the testing exposure. The relative difference in pixel sensitivity decays exponentially with the time delay, i.e., exp[-Cτ], where τ is the time delay and C is the decay constant.); and
K=O.5, which takes account the difference between light and x-ray responses of the imager and would be unity if the light were replaced by x-rays. This factor can be derived numerically (once C is known) by searching its optimum value to correct for a known ghosting image using the current procedure (For example, right after the ghosting exposure above, take step 2. Then follow steps 3-4 to correct for the ghosting in the testing images using K as an adjustable parameter. Adjust the value of K and choose the value that gives the best result); 4. Take the Nth x-ray image (uncorrected) at time t (with the patient in the x-ray beam) to derive an image called Nth X-ray Image
uncoriected (t) and use G
gh
0Sting(t) to derive a ghost-corrected image called Nth X-ray Image
coπected (t) according to: Nth X-ray Image
co,τected (t) = Nth X-ray Image
uncoπ . ected (t)/G
ghostin
g(t);
5. Repeat steps 2-4 for the next x-ray portal image acquisition; and 6. In case when there is still a small but long lasting residual ghosting signal in the corrected images, re-do the initial calibration, i.e., repeat steps 0-1 before repeating steps 2-4. A particular benefit of the preferred and exemplary embodiments described in this patent specification is that the information for making the ghost correction can be derived, and the ghosting correction can be made, rapidly and while the patient is resting on the treatment table. The pertinent steps needed to implement the ghosting correction can take place in short intervals between exposures to high-energy treatment radiation for portal imaging or for treatment. Because the preferred embodiment takes into account the length of the time interval from time t
0 to time t, it
can properly take into consideration changes in sensitivity over that time interval, and can accommodate time intervals of durations suitable to particular radiation treatment equipment or medical preference. In addition to any known corrections, such as a correction for dark current effects as well as a correction for any initial gain variation among different pixels as in known in the art of using an imager of the type identified above (see step 0 above), corrections are made for ghosting effects that may occur due to prior exposure to high-energy radiation. To this end, as illustrated in Fig. 6, at step 60 the radiation treatment system that embodies the improvement disclosed here takes a light image by flashing light source 26 to thereby derive from imager 18 an image called Imageij
gi
lt_
reference- This image contains respective values for the pixel positions in the imager, and is stored in computer 20. The pixel values of this image should include contributions due to the light from source 26 in taking the image but not from ghosting. Namely, Imagejig
ht referenc
e is taken when imager 18 is well rested so that any ghost images therein have fully dissipated. This image can be taken with the patient on table 10, or at some other time. At step 62, which comes in time after a portal x-ray image has been taken with the patient on table 10, say the (N-l)th x-ray image, where N= 1,2,3,..., another light image is taken at time t
0 and is stored in computer 20. This image is called Image(
N- i)th_iight testi
ng(t
0), and can be taken without moving the patient or imager 18. It also includes respective values for the pixel positions. The pixel values of this image should include contributions due to the light from source 26 in taking the image and
from any ghosting remaining in the imager due to x-ray exposures through the (N-I) exposure. At step 64, computer 20 calculates a gain correction G
ghostmg(t) for a time t, at which the next (i.e., Nth) x-ray image will be taken. Fig. 7 illustrates the time difference (t-to) in relation to the times of taking the x-ray images (N-I) and N, and the talcing the light images that follow the (N-l)th and the Nth x-ray images, respectively. The gain correction G
ghOstmg(t) also comprises respective values for the pixel positions, calculated according to: G
gl
1Osting(t)=l+ K* {
lighUestingCto)/ Imageiightj-efeience] | noimalized to 1 at a iefeience pixel "1 }
exp["C(t-to)] where K=O.5, which takes account the difference between light and x-ray responses of the imager and would be unity if the light were replaced by x- rays; and C — a known constant, determined experimentally by testing imager 18 to determine a rate of decay of a ghost image due to exposure to high- energy radiation.
At step 66, with the patient still on table 10, the Nth x-ray image is taken at time t to derive an image called Nth X-ray Itnageuncorrected (t). This image also is on a pixel-by-pixel basis and is stored in computer 20. It may contain ghosting, and is not yet corrected for ghosting. At step 68, computer 20, with suitable software, calculates an x-ray portal image that is corrected for ghosting, called Nth X-ray ImageCOπected (t), according to:
Nth X-ray Imagecorrected (t) = Nth X-ray Imageuncoπected (t)/Gghostmg(t). Steps 62-68 can be repeated for the next x-ray portal image acquisition. The portal x-ray images that have been corrected for ghosting as described above can be used in the manner known in the art to guide radiation treatment. More generally, this patent specification describes a method and an apparatus for reducing ghosting in a flat panel imager that images high-energy radiation passing through a patient, by means of exposing the imager to radiation different from the high-energy radiation and using image information derived in response to such different radiation to correct image information derived in response to high-energy radiation so as to reduce undesirable ghosting effects. More specifically, the new method and apparatus involve using such corrections with imagers that directly convert x-ray energy to electronic signals rather than first converting x-ray energy to light energy. Still more specifically, the novel method and apparatus involve conversion utilizing a Selenium-based layer. The different radiation used in the correction can be light of a wavelength or wavelengths, or one or more ranges of wavelengths, that produce a suitable imaging output from the imager. A method for correcting for ghosting in portal imaging with high-energy radiation and a digital flat panel imager, according to another exemplary embodiment, will now be discussed with reference to Fig. 8. The digital flat panel imager is exposed to light to derive a light reference image (step S81). The imager is exposed to high energy radiation before time to. A light testing image is derived by exposing the digital flat panel imager to light at time t0 (step S83). A gain correction is calculated for a portal image to be taken at a time t, using the light reference image,
the light testing image, and a factor accounting for changes in sensitivity or gain of the imager in an interval between the times to and t as well as a factor accounting for different response or sensitivity of the imager to light from that to x-rays (step S85). When the digital flat panel imager is later exposed to high-energy radiation at time t to derive the portal image (step S87); the gain correction is used to correct the portal image for ghosting due to the exposing the imager to high-energy radiation before the exposure at the time t (step S89). Exposing the imager to high-energy radiation to derive the portal image can comprise a direct conversion to electronic output from the imager. At least one of the exposures to light to derive the light reference image and the light testing image may comprise exposure to a flat field light. According to another exemplary embodiment (Fig. 9), a radiation treatment method comprises (i) deriving a portal image by exposing a patient to high-energy radiation and imaging the radiation with a digital flat panel imager (step S91), (ii) correcting the portal image for ghosting due to previous exposure to high-energy radiation by modifying the portal image in accordance with information regarding a response of the imager to light and at least one of a factor related to decay of images of high-energy radiation taken with the imager and a factor related to different responses of the imager to light and to high-energy radiation, to thereby derive a corrected portal image (step S93), and (in) using the corrected portal image to guide radiation treatment of the patient (step S95). Deriving a portal image may comprise using a direct conversion imager to derive the portal image. Using the corrected portal image may comprise selecting a relative position between a patient and a source of high-energy radiation in accordance with the corrected portal image. The
method may further comprise correcting in accordance with a light image taken with the imager when the imager is substantially free of ghosting due to exposure to high- energy radiation.
One example of software carrying out the steps illustrated in Fig. 6 and
discussed above is reproduced below. The software is written in C and can be run on
a computer using a Windows NT operating system. Of course, persons of ordinary
skill in the relevant technology can adapt the program listing below, in light of the
teaching above, to different computer languages and different operating systems.
The relationship between the notation above and that in the program reproduced
below is as follows:
Imageiightjeference = Init.img
Image(N-l )thjight_testing(tθ) = Test.img
t ≡ tl
to ≡ to Nth X-ray Imageunc0ITected (t) ≡ Uncor.img
Nth X-ray ImageC01Tected (t) ≡ Outputtif
#include <stdlib.h> #include <stdio.h> #include <math.h>
#include "gtiff.h"
// functions int ImageCorrect () ; int get_params () ; int init () ; FILE *OpenImageFile (char *fname) ; void calculate_k() ; void Error(char *msg) ;
// variables
unsigned long FileBuffer = 10241*10241;
double tl, // time variable t0=0.0, // reference time
CC=O.395, // C constant fk; // normalizing constant
// Image size int width=2560, height=3072, size; // image size in pixels
// Image array pointer unsigned short *img_out;
// Image File names char img_t_Fname [_MAX_PATH] ="c:\\ghosting\\Uncor. img" , img_test_Fname [_MAX_PATH] ="c:\\ghoεϊting\\Test.img" , img__init__Fname [_MAX_PATH] ="c:\\ghosting\\lnit.img" ,
img_out_Fname [_MAX_PATH] ="c:\\ghosting\\Output.tif";
// Image File pointers FILE *img__t_fp, *img_test_fp, *img_init_fp;
main()
// create image get_params () ; init () ; ImageCorrect () ;
// tiff save TIFF_SaveToFile (img_out_Fname, (char *)img_out, width, height, 16, 0) ;
// release memory free (img_out) ;
// close files if(img_t_fp) fclose (img_t_fp) ; if (img_test_fp) fclose (img_test_fp) ; if (img_init_fp) fclose (iτng_init_fp) ;
// wrap up fprintf (stderr, "\nDone !\n\a");
return 0; }
int get_params ( ) {
// filenames if ( ! (*img_init_Fname) ) { fprintf (stderr, "\nlmage (init) file name: ") ; scanf ("%s" , img_init_Fname) ; }
if ( ! (*img_test_Fname) ) { fprintf (stderr, "\nlmage (testing) file name: ") ; scanf ("%s", img_test_Fname) ; }
if ( ! (*img_t_Fname) ) { fprintf (stderr, "\nlmage(t) file name: ") ; scanf ("%s", img_t_Fname) ; }
if ( ! (*img_out_Fname) ) { fprintf (stderr, "\nlmage (output) file name: ") ; scanf ("%s" , img_out_Fname) ,■ }
// time variable if(!tθ) { fprintf (stderr, "\nreference time (to) : ") ; scanf ("%lf", &tθ) ; } if(iti) { fprintf (stderr, "\ntime: ") ; scanf ("%lf" , &tl) ; }
return 0; }
int ImageCorrect () { int i; double gain, v; unsigned short tst, in, img; unsigned short *p = img_out;
double E = exp(-CC* (tl-tθ) ) ;
for(i=0; i<size; i++, p++) {
// read pixel values from images; fread(&tst, 2, 1, img_test_fp) / fread(&in, 2, 1, img_init_fp) ; fread(&img, 2, 1, img_t_fp) ;
// calculate corrected pixel value if (in != 0) { gain = 1.0 + 0.5* (fk * 1.0) * E; v = (double) img / gain; }
else v = (double) img;
'if (v>0 && v< 65535) { *p = (unsigned short)v; }
else *p=img; }
return 0; }
int init () { // open input image files if ( ! (img_init_fp=OpenImageFile (img__init_Fname) ) ) { Error( "Incorrect type or size of file buffer for Image (init) ") ; }
if ( ! (img_test_fp=OpenImageFile (img__test_Fname) ) ) { Error( "Incorrect type or size of file buffer for Image (testing) ") ; }
if ( ! (img_t_fp=OpenImageFile (img_t_Fname) ) ) { Error( "Incorrect type or size of file buffer for Image (t) ") ; ' }
// calculate image size in pixels size = width*height;
// allocate output image buffer img_out = (unsigned short *) calloc (size, sizeof (short) ); if (!img_out) { Error("Could not allocate memory for output image") ; }
// calculate normalizing constant calculate_k() ;
return 0;
void calculate_k() { long offs; unsigned short tst, in;
// find the file offset of center pixel offs = (long)width* ( (long) height-1) -2;
// get corresponding pixel values from files fseek(img_test_fp, offs, SEEK__SET) ; fread(&tst, 2, 1, img_test_fp) ;
fseek(img_init_fp, offs, SEEK_SET) ; fread(&in, 2, 1, img_init_fp) ;
// error situation (division by 0) if(!tst) Error("Center pixel in test image is zero") ;
// calculate k fk = (double) in/ (double) tst;
// restore file positions to beginning of files fseek(img_test_fp, 0, SEEK_SET) ; fseek(img_init_fp, 0, SEEK_SET) ;
}
FILE *OpenImageFile (char *fname) { FILE *fp;
if (! (fp=fopen(fname, "rb"))) return NULL; if (setvbuf (fp, NULL, _IOFBF, FileBuffer) ) { fclose(fp); return NULL; } return fp; }
void Error(char *msg) { fprintf ( stderr , " \nERR0R * * * * %s\n\a" , msg) ; exit ( l) ; }
Elements and/or features of different illustrative embodiments may be
combined with each other and/or substituted for each other within the scope of this disclosure and appended claims. The above specific embodiments are illustrative, and many variations can be introduced on these embodiments without departing from the spirit of the disclosure or from the scope of the appended claims. This application claims the priority of U.S. provisional application Serial No. 60/454,988, filed March 14, 2003 and entitled "PORTAL IMAGING OF HIGH- ENERGY RADIATION WITH AN ELECTRONIC FLAT PANEL IMAGER", which is incorporated herein by reference.